R help - Jan 2009 - Fitting pareto to some data

Dear R-users,

I am trying to fit pareto distribution to some data but i've one problem.
Using optim to calculate the maximum of the likelihood function of the
pareto I use as start parameters the moments method(using the distribution
function in the package actuar):

media=mean(x)
var=mean(x^2)-media^2
scale=2*var/(var-media^2)
shape=(scale-1)*media

fitdistr(x,dpareto,list(shape=shape,scale=scale))

The problem is how to initialize the shape and scale parameters when
"var<=media^2" . When "var<=media^2" scale and shape
parameters are negative
using moments method.
Does anyone know how to initialize the paramaters when
"var<=media^2"?

Thanks
Borja

	[[alternative HTML version deleted]]